一种软阈值滤波算法的改进  

An Improved Threshold Filtering Algorithm Based on Wavelet

作　　者：陈利霞[1] 丁宣浩[1] 王东[1] 

机构地区：[1]桂林电子工业学院计算科学与应用物理系,广西桂林541004

出　　处：《桂林电子工业学院学报》2003年第3期10-13,共4页Journal of Guilin Institute of Electronic Technology

基　　金：广西自然科学基金项目(0135007);广西科技攻关项目(02280204)

摘　　要：由于各尺度上小波系数的微小的偏移对相关系数有影响,为了得到更精确的噪声方差,文中利用区域相关系数对王博等人提出的相关算法进行改进,得到新的软阈值。最后通过仿真,将此阈值与软阈值manimaxi和heursure进行比较,结果表明利用改进后的软阈值去噪效果较佳。As very slight excursion of the wavelet coefficient may lead to incorrect correlated coefficient in each scale and in order to get more accurate variance of noise,the authors used the relativity of the wavelet coefficients in region to improve the correlative algorithm proposed by Wang Bo and other authors and gained a new soft threshold.Compared with the soft threshold minimaxi and heursure,the simulation showed that the modified algorithm is superior to its predecessor in terms of filtering function.

关 键 词：小波变化 软阈值 区域相关系数 多分辨分析 去噪算法 

分 类 号：TN911.4[电子电信—通信与信息系统]